Video Processing for AI Agents

Rendobar is the only production MCP-native video processing API. Connect any MCP-compatible AI agent and process media through natural language: caption, compress, compose, generate images, and run any FFmpeg command.

Connect your agent — $5 free View docs →

What is MCP?

MCP video tools let AI agents process media files through a standardized protocol. Model Context Protocol (MCP) is an open standard by Anthropic that gives AI agents a consistent way to discover and use external tools. Instead of building custom integrations for each agent, an MCP server exposes a set of tools that any compatible client can call automatically. Rendobar runs a hosted MCP server with 10 tools covering the full media processing lifecycle, from uploading files to downloading finished results. There is also a local server, the @rendobar/mcp npm package (source on GitHub), which runs on your machine so an agent can upload files straight off your disk.

This means an AI agent like Claude can process video without writing code. A user says "burn captions into this video," and the agent calls submit_job with the right parameters, polls get_job for completion, and returns the download link. No scripts, no deployment pipelines, no infrastructure.

10 tools, full media control

Every MCP tool maps to a REST API endpoint. Same authentication, same billing, same capabilities. Your agent discovers these tools automatically on connection.

submit_job

Submit any media processing job. Burn or animate captions, compress to a target size, compose a timeline, generate, edit or upscale images, inspect media, or run any FFmpeg command. Returns a job ID for tracking.

get_job

Retrieve the status, progress, and results of a submitted job. Returns download URLs when the job completes.

list_jobs

List recent jobs with filters for status, type, and date range. Paginated results with metadata.

probe_media

Inspect a media file without processing it. Returns duration, resolution, codecs, bitrate, frame rate, and audio channel layout.

get_account

Check the current credit balance, active plan, usage limits, and remaining quota. Useful before submitting expensive jobs.

cancel_job

Cancel a job that is waiting, dispatched, or already running. Credits for unfinished work are not charged.

upload_media

Start an upload session for local files. Agents with shell access get presigned URLs, everyone else gets a link page that works from a phone.

complete_upload

Finalize an upload once its bytes have landed in storage, turning it into an asset a job can reference.

get_upload

Check an upload session's progress. Supports a wait mode so the agent can block instead of polling.

list_job_types

Read the live job registry: every active job type, what it is for, and how its output chains into another job.

Connect in under a minute

Add Rendobar to your Claude Desktop configuration. Replace the API key with your own from the dashboard.

// Add to your Claude Desktop config
{
  "mcpServers": {
    "rendobar": {
      "url": "https://api.rendobar.com/mcp",
      "headers": {
        "Authorization": "Bearer rb_your_api_key"
      }
    }
  }
}

User:

Burn captions into this video: https://cdn.example.com/demo.mp4

Agent:

I'll submit a caption burn job for your video.
submit_job({ type: "caption.burn", inputs: { source: "..." }, params: { fontSize: 42 } })

Job job_abc123 submitted. Checking status...

Done. Download your captioned video:
https://cdn.rendobar.com/jobs/job_abc123/output.mp4

Works with every MCP client

Claude Desktop

Native MCP support. Add the server config and start processing videos from conversation.

Claude Code

Use MCP tools directly from the CLI. Automate video pipelines in development workflows.

GPT (via MCP bridge)

Connect through an MCP-to-function-calling bridge. Same tools, same API key.

Any MCP-compatible agent

The MCP standard is open. Any agent that speaks the protocol can use Rendobar's tools.

How it works

  1. Connect your agent

    Add Rendobar as an MCP server in your agent's configuration. Point it to api.rendobar.com/mcp with your API key. One-time setup, under a minute.

  2. Agent calls tools

    Your AI agent discovers available tools automatically. It can submit jobs, check status, probe media files, and manage uploads — all through natural language instructions.

  3. Get results

    Jobs run on Rendobar's serverless infrastructure. The agent polls for completion and retrieves download URLs. No servers, no queues, no infrastructure to manage.

Or let your coding assistant integrate it

Paste one line into Claude, ChatGPT, or Cursor and the assistant reads the full integration playbook, detects your stack, and wires Rendobar into your codebase. Your API key stays in your env file, never in the chat.

Add Rendobar to my app: read and follow https://rendobar.com/prompts/integrate.md

The playbook lives at rendobar.com/prompts/integrate.md and stays current with the platform.

What agents build with MCP video tools

MCP turns video processing from a developer task into a conversational one. These are real workflows that agents handle end-to-end.

Social cutdown pipelines

An agent takes one long recording, cuts it to a vertical aspect ratio, burns captions, and compresses each cut to the size budget its platform accepts. The whole pipeline is orchestrated through MCP calls, with no backend code required.

Automated caption workflows

A content team asks their agent to "add captions to all videos in this folder." The agent uploads each file via upload_media, submits caption extraction jobs, reviews the generated subtitles, and burns them into the final output. Handles dozens of files in a single conversation.

Brand-compliant video preparation

Marketing teams instruct their agent to compress each cut to a platform size budget and burn on-brand captions. The agent probes each file first with probe_media to verify resolution and codec, then submits the appropriate jobs.

Developer workflow automation

Developers use Claude Code to process test videos during development. "Transcode this to WebM at 720p" or "extract the first 10 seconds" — the agent calls submit_job with FFmpeg parameters and returns the result, all without leaving the terminal.

Same pricing, same credits

MCP tools use the same credit-based billing as the REST API. Every account starts with $5 in free credits — no credit card required. Jobs submitted via MCP cost exactly the same as jobs submitted via REST. The Pro plan ($9/month) includes $5/month in credits, 20% bonus on purchases, and 20 concurrent jobs. View full pricing →

Frequently asked questions

What is MCP?

Model Context Protocol (MCP) is an open standard created by Anthropic that defines how AI agents discover and use external tools. It replaces custom function-calling integrations with a single, consistent protocol. An MCP server exposes tools that any compatible agent can call — no per-agent integration code required.

Which AI agents support MCP?

Claude Desktop and Claude Code support MCP natively. Other agents like GPT-based assistants can connect via MCP-to-function-calling bridges. Any agent framework that implements the MCP client specification can use Rendobar's tools directly.

What video operations can I do via MCP?

Every operation available through the REST API is also available via MCP. That includes burning and animating captions, compressing to a target size, composing a timeline, generating, editing and upscaling images, inspecting media with ffprobe, and running any FFmpeg command.

Do I need an API key for MCP?

Yes. MCP uses the same API key authentication as the REST API. Generate a key from the Rendobar dashboard and include it in your MCP server configuration. The same credit-based billing applies — no separate MCP pricing.

Is MCP access free?

Every account starts with $5 in free credits. MCP tools consume credits at the same rates as equivalent REST API calls. No credit card required to start. You can purchase additional credits when your balance runs low.

Can I process multiple videos in one conversation?

Yes. Your agent can call submit_job for each video and use list_jobs to track all active jobs. There is no per-conversation limit and no submission cap. Jobs run at your plan's concurrency (1 on Free, 20 on Pro) and any extra wait in a queue, starting automatically as slots free.

How is MCP different from the REST API?

MCP and REST are two interfaces to the same platform. The REST API is for programmatic integrations — scripts, backends, CI pipelines. MCP is for AI agents — your agent discovers tools, decides which to call, and handles the results. Same jobs, same billing, same infrastructure behind both.

Related

Connect your agent today

$5 free credits on signup. No credit card. 10 tools, full media processing stack.

Get started free Read the docs →

Recent changes

All updates